e r l a n g : c o o k b o o k

This Web

All Webs

Erlang Links

Erlang Web Ring

Testing Similarity of Two Strings via Levenshtein Distance

Problem

You want to determine how "similar" two strings (or lists, or vectors) are, such as for spelling correction or pattern-matching.

Solution

Note: This library does not exist yet. Scheme data shown for the time being:

One metric of string similarity is the Levenshtein Distance, or edit distance, which is a count of character insertions, deletions, and substitutions that will transform one string to another. To get the Levenshtein Distance of two strings, use the levenshtein.scm library (available from http://www.neilvandyke.org/levenshtein-scm/) and load it like:

(require (lib "levenshtein.ss" "levenshtein"))

Then you can use the string-levenshtein procedure:

> (string-levenshtein "adresse" "address")
2

You can also get the Levenshtein Distance of combinations of strings, lists, and vectors. For example:

> (levenshtein '#(#\a #\d #\r #\e #\s #\s #\e) "address")
2

See the levenshtein.scm documentation for additional procedures that can be used to improve performance in some applications.

Discussion

Comments about this recipe

Contributors

-- NeilVanDyke - 16 May 2004

Discussion

This doesn't apply to Erlang, and is only here as a placeholder until the library is implemented. Coming to a Jungerl near you...

-- BrentAFulgham - 23 Aug 2004

You may want to take a look at my implementation...

-- AdamLindberg - 21 Dec 2005